Port presence feature forward from orphaned agent/dd2a126b57a1 branch

Recovers a real August feature that was never merged (3 WIP checkpoints +
fix commit 47ab492, 225 commits behind main): a Redis-backed heartbeat
presence store plus /api/presence/{heartbeat,presence,status} endpoints.
Ported onto current main's conventions rather than rebuilt:

- app/models/user.py: last_seen_at/status_message columns (unchanged from
  the orphaned branch; still compatible with the current model).
- alembic/versions/0006_add_presence_columns.py: same column migration,
  rebased as the new head after 0005_add_world_content_tables (was
  2a9b9873244a off 0001_initial).
- app/routers/presence.py: same three endpoints, updated to the DbDep
  pattern other routers (avatar.py) now use; retains the 47ab492 fix
  (is_online_batch via MGET, correct StatusUpdate Field annotation).
- app/services/presence.py: real adaptation beyond a mechanical port --
  the original PresenceStore required an explicit connect() that nothing
  ever called (main.py's lifespan never wired it up), so every real
  request would have raised "not connected". Made the Redis client connect
  lazily on first use instead, and cache it per running event loop rather
  than globally: this repo's TestClient(app) pattern (no `with` block) can
  dispatch different requests onto different event loops, which broke a
  single cached connection the same way NullPool's comment in conftest.py
  describes for asyncpg. Dropped the unused get_presence()/
  update_status_message() methods and the never-wired init/shutdown
  singleton helpers -- dead code even in the original.
- tests/conftest.py: extended the existing ephemeral-Postgres bootstrap to
  also bring up ephemeral Redis (docker-compose.yml already defined the
  service) when REDIS_URL isn't already reachable, so `pytest` stays
  hermetic.
- tests/test_presence.py: heartbeat sets online, batch presence over
  mixed online/offline/unknown ids (the exact 47ab492 regression), status
  message update/strip/length-limit, and TTL expiry via a short-TTL
  PresenceStore instance (real Redis SET/EXISTS/MGET, no 60s sleep needed).

Verified: full backend pytest suite (144 tests) passes against a fresh
Postgres+Redis. Also drove the real endpoints end-to-end against a running
uvicorn + docker-compose Postgres/Redis: heartbeat -> online, batch query,
forced TTL expiry -> offline with last_seen_at retained, status message
trim + 422 over 140 chars.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y2FESCaMXeJpC9gSctiEeD

backend/app/main.py

@@ -6,7 +6,7 @@ from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 
 from app.config import settings
-from app.routers import auth, users, worlds, assets, avatar, world_content, internal
+from app.routers import auth, users, worlds, assets, avatar, world_content, internal, presence
 
 
 from app.services.game_servers import supervisor
@@ -44,6 +44,7 @@ app.include_router(world_content.router, prefix="/api/worlds", tags=["world-cont
 app.include_router(assets.router, prefix="/api/assets", tags=["assets"])
 app.include_router(avatar.router, prefix="/api/avatar", tags=["avatar"])
 app.include_router(internal.router, prefix="/api/internal", tags=["internal"])
+app.include_router(presence.router, prefix="/api/presence", tags=["presence"])
 
 
 @app.get("/api/health")

backend/app/models/user.py

@@ -1,7 +1,7 @@
 from datetime import date
 from typing import Optional
 
-from sqlalchemy import Column, Integer, String, Boolean, DateTime, Date
+from sqlalchemy import Column, Integer, String, Boolean, DateTime, Date, Text
 from sqlalchemy.orm import relationship
 from sqlalchemy.sql import func
 
@@ -20,6 +20,10 @@ class User(Base):
     created_at = Column(DateTime(timezone=True), server_default=func.now())
     updated_at = Column(DateTime(timezone=True), onupdate=func.now())
 
+    # Presence / status (backend/app/services/presence.py, app/routers/presence.py)
+    last_seen_at = Column(DateTime(timezone=True), nullable=True)
+    status_message = Column(Text, nullable=True)
+
     sessions = relationship("Session", back_populates="user", cascade="all, delete-orphan")
     refresh_tokens = relationship("RefreshToken", back_populates="user", cascade="all, delete-orphan")
     worlds = relationship("World", back_populates="owner")

backend/app/routers/presence.py

@@ -0,0 +1,146 @@
+"""Presence & status router.
+
+Endpoints:
+- POST  /api/presence/heartbeat  -- record a heartbeat (sets Redis TTL key)
+- GET   /api/presence/presence   -- query presence for a list of user ids
+- PATCH /api/presence/status     -- update the authenticated user's status message
+"""
+
+from datetime import datetime, timezone
+from typing import Annotated
+
+from fastapi import APIRouter, HTTPException, status, Depends
+from pydantic import BaseModel, Field, field_validator
+from sqlalchemy import select
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from app.database import get_db
+from app.dependencies import CurrentUser
+from app.models.user import User
+from app.services.presence import presence_store
+
+router = APIRouter()
+
+DbDep = Annotated[AsyncSession, Depends(get_db)]
+
+# ──────────────────────────── Pydantic schemas ────────────────────────────────
+
+_STATUS_MAX_LEN = 140
+
+
+class HeartbeatRequest(BaseModel):
+    """Empty body -- the caller's identity comes from the JWT."""
+
+
+class PresenceUser(BaseModel):
+    user_id: int
+    is_online: bool
+    last_seen_at: datetime | None
+    status_message: str | None
+
+
+class PresenceResponse(BaseModel):
+    users: list[PresenceUser]
+
+
+class StatusUpdate(BaseModel):
+    status_message: str | None = Field(default=None, max_length=_STATUS_MAX_LEN)
+
+    @field_validator("status_message")
+    @classmethod
+    def strip_message(cls, v: str | None) -> str | None:
+        if v is not None:
+            v = v.strip()
+        return v
+
+
+# ──────────────────────────────── Endpoints ───────────────────────────────────
+
+
+@router.post("/heartbeat", response_model=dict)
+async def heartbeat(
+    _req: HeartbeatRequest,
+    current_user: CurrentUser,
+    db: DbDep,
+):
+    """Record a heartbeat for the authenticated user.
+
+    Sets a Redis key with a TTL so the server can derive online/offline
+    status without trusting client claims. Also persists last_seen_at to
+    the database.
+    """
+    # Update Redis TTL (presence heartbeat)
+    await presence_store.heartbeat(current_user.id)
+
+    # Persist last_seen_at to the database
+    now = datetime.now(timezone.utc)
+    result = await db.execute(select(User).where(User.id == current_user.id))
+    user = result.scalar_one_or_none()
+    if user:
+        user.last_seen_at = now
+        await db.commit()
+
+    return {"status": "ok", "user_id": current_user.id}
+
+
+@router.get("/presence", response_model=PresenceResponse)
+async def get_presence(
+    user_ids: str,
+    current_user: CurrentUser,
+    db: DbDep,
+):
+    """Return presence info for a comma-separated list of user ids.
+
+    Online/offline is derived from the Redis heartbeat TTL.
+    last_seen_at and status_message come from the database.
+    """
+    ids = [int(uid.strip()) for uid in user_ids.split(",") if uid.strip()]
+    if not ids:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="user_ids must contain at least one integer",
+        )
+
+    # Check Redis for online status (single round trip via MGET).
+    online_ids = await presence_store.is_online_batch(ids)
+
+    # Fetch DB records for last_seen_at and status_message
+    result = await db.execute(select(User).where(User.id.in_(ids)))
+    users = {u.id: u for u in result.scalars().all()}
+
+    presence_users = []
+    for uid in ids:
+        user = users.get(uid)
+        if user is None:
+            continue
+        presence_users.append(
+            PresenceUser(
+                user_id=uid,
+                is_online=uid in online_ids,
+                last_seen_at=user.last_seen_at,
+                status_message=user.status_message,
+            )
+        )
+
+    return PresenceResponse(users=presence_users)
+
+
+@router.patch("/status", response_model=dict)
+async def update_status(
+    body: StatusUpdate,
+    current_user: CurrentUser,
+    db: DbDep,
+):
+    """Update the authenticated user's status message (length-limited)."""
+    result = await db.execute(select(User).where(User.id == current_user.id))
+    user = result.scalar_one_or_none()
+    if not user:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail="User not found",
+        )
+
+    user.status_message = body.status_message
+    await db.commit()
+
+    return {"status": "ok", "status_message": body.status_message}

backend/app/services/presence.py

@@ -0,0 +1,145 @@
+"""Redis-backed presence store.
+
+Presence is determined by a heartbeat TTL key in Redis:
+- heartbeat(user_id) sets a TTL key per user (default 60 s).
+- is_online(user_id) / is_online_batch(user_ids) check whether the key(s)
+  exist.
+
+status_message and last_seen_at are persisted in the ``users`` table by the
+router (app/routers/presence.py); this store only owns the Redis side.
+
+The Redis client is created lazily, on first actual use, rather than
+requiring an explicit ``connect()`` call during app startup. This matters
+because FastAPI's lifespan handler does not run under the plain
+``TestClient(app)`` instantiation this codebase's tests use (no
+``with`` block), so anything that only connected during lifespan startup
+would never connect in tests. Lazy connection means the store works
+identically whether the app was started for real or is being driven by
+pytest.
+
+The client is also cached per-*event loop* rather than globally: the same
+plain ``TestClient(app)`` pattern can dispatch different requests onto
+different event loops (see the NullPool comment on ``_override_db`` in
+tests/conftest.py for the analogous asyncpg problem), and an
+``aioredis.Redis`` connection, like an asyncpg connection, breaks with
+"Event loop is closed" if reused across loops. In the real running server
+there is exactly one event loop for the process lifetime, so this
+degrades to a single cached, pooled connection there.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+
+import redis.asyncio as aioredis
+
+from app.config import settings
+
+logger = logging.getLogger(__name__)
+
+# Key prefix for heartbeat entries.
+_PRESENCE_KEY_PREFIX = "presence:heartbeat:"
+
+# Default heartbeat TTL in seconds (1 minute).
+# Clients should heartbeat at least once per interval.
+_DEFAULT_HEARTBEAT_TTL = 60
+
+
+class PresenceStore:
+    """Redis-backed presence store with async, lazily-connected operations."""
+
+    def __init__(self, redis_url: str | None = None, heartbeat_ttl: int = _DEFAULT_HEARTBEAT_TTL):
+        # redis_url is resolved from settings at first use (not at
+        # construction time) so tests that override settings.REDIS_URL
+        # before the first call still take effect -- this module is
+        # imported (and the module-level singleton below constructed)
+        # well before any such override runs.
+        self._redis_url = redis_url
+        self._heartbeat_ttl = heartbeat_ttl
+        # Keyed by id(running loop) -- see the module docstring.
+        self._clients: dict[int, aioredis.Redis] = {}
+        self._connect_lock = asyncio.Lock()
+
+    # ------------------------------------------------------------------
+    # Lifecycle
+    # ------------------------------------------------------------------
+
+    async def _ensure_client(self) -> aioredis.Redis:
+        """Return a connected client for the current event loop, connecting
+        on first use per loop."""
+        loop_key = id(asyncio.get_running_loop())
+        client = self._clients.get(loop_key)
+        if client is None:
+            async with self._connect_lock:
+                client = self._clients.get(loop_key)
+                if client is None:
+                    url = self._redis_url or settings.REDIS_URL
+                    client = aioredis.from_url(
+                        url,
+                        encoding="utf-8",
+                        decode_responses=True,
+                    )
+                    try:
+                        await client.ping()
+                    except Exception:
+                        logger.warning("Redis ping failed for %s", url)
+                        raise
+                    self._clients[loop_key] = client
+        return client
+
+    async def close(self) -> None:
+        """Close the Redis connection for the current event loop, if one was
+        opened."""
+        loop_key = id(asyncio.get_running_loop())
+        client = self._clients.pop(loop_key, None)
+        if client is not None:
+            await client.aclose()
+
+    # ------------------------------------------------------------------
+    # Heartbeat
+    # ------------------------------------------------------------------
+
+    async def heartbeat(self, user_id: int) -> None:
+        """Record a heartbeat for *user_id*.
+
+        Sets a TTL key so that the user is considered online as long as
+        heartbeats arrive at least once per heartbeat_ttl seconds.
+        """
+        client = await self._ensure_client()
+        await client.set(self._heartbeat_key(user_id), "1", ex=self._heartbeat_ttl)
+
+    # ------------------------------------------------------------------
+    # Online checks
+    # ------------------------------------------------------------------
+
+    async def is_online(self, user_id: int) -> bool:
+        """Return True if the user has a valid heartbeat key."""
+        client = await self._ensure_client()
+        exists = await client.exists(self._heartbeat_key(user_id))
+        return bool(exists)
+
+    async def is_online_batch(self, user_ids: list[int]) -> set[int]:
+        """Return the subset of user_ids that currently have a valid
+        heartbeat key. Uses MGET (one round trip) rather than one EXISTS
+        per user.
+        """
+        if not user_ids:
+            return set()
+        client = await self._ensure_client()
+        keys = [self._heartbeat_key(uid) for uid in user_ids]
+        values = await client.mget(keys)
+        return {uid for uid, v in zip(user_ids, values) if v is not None}
+
+    # ------------------------------------------------------------------
+    # Helpers
+    # ------------------------------------------------------------------
+
+    @staticmethod
+    def _heartbeat_key(user_id: int) -> str:
+        return f"{_PRESENCE_KEY_PREFIX}{user_id}"
+
+
+# Singleton instance shared by the router, so all requests share the same
+# connection pool.
+presence_store = PresenceStore()